-
-
Interface
Inspiration
Most resume tools today either hide behind vague ATS scores or rely entirely on black-box AI that never explains why a resume fails. As students and early developers, we experienced how confusing and discouraging this process can be.
We wanted to build something honest, explainable, and engaging — a tool that doesn’t just judge resumes, but helps users grow. That idea became Resume Roaster AI, where users choose how much truth they’re ready to face.
What it does
Resume Roaster AI analyzes resumes against a specific job role and:
Scores resumes using real NLP and Machine Learning
Identifies missing or weak skills
Provides feedback in three modes:
💚 Soft – encouraging and polite
🟠 Constructive – honest with clear improvements
🔴 Brutal – unfiltered resume roast
Highlights gaps visually
Offers a recovery mode — “Want a Hug?” — that gives actionable steps to improve the resume
It turns resume feedback into an interactive, transparent experience.
How we built it
PDF Parsing to extract resume text
NLP preprocessing (cleaning, tokenization, stopword removal)
TF-IDF vectorization to represent resume and job role text
Cosine similarity to compute an explainable match score
Flask backend to handle analysis
Dark neon frontend with glassmorphism UI
GSAP animations for smooth interactions
The system is designed to work without depending entirely on generative AI, ensuring reliability and explainability.
Challenges we ran into
Handling inconsistent resume PDF formats
Avoiding over-reliance on LLMs while keeping feedback engaging
Designing feedback that is brutal but still useful
Maintaining explainability in ML scoring
Balancing strong UI design with limited hackathon time
Each challenge helped refine both the technical and UX aspects of the project.
Accomplishments that we're proud of
Built a fully explainable resume scoring system
Designed a unique emotion-driven feedback experience
Combined serious ML with playful UX
Avoided black-box AI dependency
Delivered a polished, end-to-end product under time pressure
What we learned
How ATS systems work at a practical level
Why explainable ML matters more than flashy AI
How UX psychology affects user trust
How to integrate ML logic with real product design
How to ship a complete, production-ready project quickly
What's next for Resume Roaster AI – Dare to Face the Truth
Line-by-line resume highlighting with red-line fixes
Role-specific resume templates
Resume version comparison
Recruiter-side ATS simulation
Cloud deployment with user accounts
Log in or sign up for Devpost to join the conversation.